Research map / Vision-language and multimodal / Cross-modal learning
VENOMREC: Cross-Modal Interactive Poisoning for Targeted Promotion in Multimodal LLM Recommender Systems
Guowei Guan, Yurong Hao, Jiaming Zhang, Tiantong Wu, Fuyao Zhang, Tianxiang Chen, Longtao Huang, Cyril Leung, Wei Yang Bryan Lim
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add visual perturbation to interaction injection (ablation) | VenomRec · Clothing (Few-Shot) | ER@20: 0.2669 -> 0.4173 | Improved |
| add independent textual perturbation (ablation) | VenomRec · Clothing (Few-Shot) | ER@20: 0.4173 -> 0.5617 | Improved |
| add cross-modal interactive co-adaptation loop (CIP) (ablation) | VenomRec · Clothing (Few-Shot) | ER@20: 0.5617 -> 0.7170 | Improved |
| add cross-modal interactive co-adaptation loop (CIP) (ablation) | VenomRec · Clothing (Zero-Shot) | ER@20: 0.5881 -> 0.7317 | Improved |
| add content perturbation on top of interaction injection (Exposure Alignment alone) (ablation) | VenomRec · Clothing (Few-Shot) | ER@5: 0.0142 -> 0.1158 | Improved |
| apply AE Filter defence against the attack | VenomRec · Clothing (Few-Shot) | ER@20: 0.7170 -> 0.7174 | No clear effect |
| apply LexNorm text normalisation defence | VenomRec · Clothing (Zero-Shot) | ER@20: 0.7317 -> 0.6743 | Hurt |
| apply Freq-Rarity token filtering defence | VenomRec · Clothing (Zero-Shot) | ER@20: 0.7317 -> 0.6357 | Hurt |
| increase number of target items from 1 to 3 | VenomRec · Clothing (Few-Shot) | ER@20: 0.7170 -> 0.2904 | Hurt |
| increase number of target items from 1 to 5 | VenomRec · Clothing (Few-Shot) | ER@20: 0.7170 -> 0.1303 | Hurt |
| use attack on larger T5-base backbone instead of T5-small | VenomRec · Clothing (Few-Shot) | ER@20: 0.7170 -> 0.4850 | Hurt |
| use VenomRec instead of Shadowcast as multimodal attack | VenomRec · Clothing (Few-Shot) | ER@20: 0.0472 -> 0.7170 | Improved |
Closest papers by meaning
- MENTOR: Multi-level Self-supervised Learning for Multimodal Recommendation (AAAI 2025 · similarity 0.86)
- MTSTRec: Multimodal Time-Aligned Shared Token Recommender (ICML 2025 · similarity 0.86)
- Curriculum Conditioned Diffusion for Multimodal Recommendation (AAAI 2025 · similarity 0.86)
- Seeing Beyond Noise: Joint Graph Structure Evaluation and Denoising for Multimodal Recommendation (AAAI 2025 · similarity 0.85)
- LGMRec: Local and Global Graph Learning for Multimodal Recommendation (AAAI 2024 · similarity 0.85)
- Harnessing Multimodal Large Language Models for Multimodal Sequential Recommendation (AAAI 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
